A Regulatory Role for ADP-ribosylation Factor 6 (ARF6) in Activation of the Phagocyte NADPH Oxidase*

In activated neutrophils NADPH oxidase is regulated through various signaling intermediates, including heterotrimeric G proteins, kinases, GTPases, and phospholipases. ADP-ribosylation factor (ARF) describes a family of GTPases associated with phospholipase D (PLD) activation. PLD is implicated in NADPH oxidase activation, although it is unclear whether activation of PLD by ARF is linked to receptor-mediated oxidase activation. We explored whether ARF participates in NADPH oxidase activation by formyl-methionine-leucine-phenylalanine (fMLP) and whether this involves PLD. Using multicolor forward angle light scattering analyses to measure superoxide production in differentiated neutrophil-like PLB-985 cells, we tested enhanced green fluorescent fusion proteins of wild-type ARF1 or ARF6, or their mutant counterparts. The ARF6(Q67L) mutant defective in GTP hydrolysis caused increased superoxide production, whereas the ARF6(T27N) mutant defective in GTP binding caused diminished responses to fMLP. The ARF1 mutants had no effect on fMLP responses, and none of the ARF proteins affected phorbol 12-myristate 13-acetate-elicited oxidase activity. PLD inhibitors 1-butanol and 2,3-diphosphoglycerate, or the ARF6(N48R) mutant assumed to be defective in PLD activation, blocked fMLP-elicited oxidase activity in transfected cells. The data suggest that ARF6 but not ARF1 modulates receptor-mediated NADPH oxidase activation in a PLD-dependent mechanism. Because PMA-elicited NADPH oxidase activation also appears to be PLD-dependent, but ARF-independent, ARF6 and protein kinase C may act through distinct pathways, both involving PLD.

The phagocyte NADPH oxidase is an important innate defense system against bacterial and fungal infections. Inherited deficiencies of this enzyme result in chronic granulomatous disease, which is characterized by enhanced susceptibility to microbial infection and dysregulated inflammatory responses. Although the components of this superoxide-generating system have been the subject of intensive investigation, the signaling mechanisms responsible for oxidase activation (the respiratory burst) are complex and not clearly defined (1). Many studies indicate that several GTPases of the Ras superfamily are involved at various levels of regulation of this inflammatory process (2). Rac was identified as a third cytosolic component required for activation of the NADPH oxidase, with Rac1 as the active component in guinea pig macrophages (3) and Rac2 in human neutrophils (4). In addition, Rap1A, which localizes to the plasma membrane and granule membranes in human neutrophils, was shown to associate with cytochrome b 558 of the oxidase (5). Mutant Rap1A inhibits oxidase activity in transfected B cells (6), although its role in the system is not entirely clear.
The ADP-ribosylation factor (ARF) 1 subfamily of Ras-related proteins consists of six mammalian GTPases (ARF1-ARF6), five of which are detected in man (ARF1, 3-6)) (7). Originally identified as cofactors required for cholera toxin-catalyzed ADP-ribosylation of G␣ s (8), the ARFs have been shown to play critical roles in vesicular transport (9). Evidence supporting a role for ARF in granulocyte functions came from studies in neutrophils and HL-60 cells, where ARF1 and ARF3 were identified as cytosolic regulators of phospholipase D (PLD) (10,11). However, these studies were conducted with cell-free reconstitution assays using recombinant ARF1 or cytosol from bovine brain, and therefore the identity of the endogenous ARF(s) participating in PLD activation and subsequent phagocyte functions is not known. PLD activity and its product phosphatidic acid have been implicated in a variety of responses by stimulated phagocytes, including secretion (12,13), phagocytosis (14 -16), and activation of NADPH oxidase (17). Several other studies suggest a role for ARF in receptor-dependent signaling in phagocytes (18,19).
All human ARF mRNA species have been detected in HL-60 cells (20), and several of these ARF isoforms (ARF 1, 5, and 6) appear to activate rat brain PLD (21). The best characterized ARF protein, ARF1, is localized to the Golgi complex and is critical for vesicular transport along secretory pathways (9). Unlike ARF1, ARF6, which is the least conserved of the human ARF proteins, localizes at the cell periphery and cycles between the plasma membrane and endosomal compartments in a guanine nucleotide-dependent manner (22,23). ARF6 was characterized as a regulator of membrane trafficking (22,23) and remodeling of the plasma membrane and the underlying cytoskeleton (24 -26). ARF6 has been linked functionally to PLD activation. In several mammalian cells ARF6(T27N) colocalizes with hPLD1a and hPLD1b (27), whereas in chromaffin cells ARF6 appears to activate PLD in vivo during exocytosis (28). Both ARF6 and PLD have been implicated in cells undergoing phagocytosis. ARF6 mutants defective in GTP binding (T27N) or GTP hydrolysis (Q67L) inhibit Fc␥ receptor (Fc␥R)-mediated phagocytosis in the RAW 264.7 macrophage cell line (29). Phagocytosis of complement-opsonized particles activates PLD in macrophages. Using Mycobacterium tuberculosis, Kusner and colleagues (14) correlated inhibition of phagocytosis with diminished PLD activity and demonstrated that phagocytosis could be restored by exogenous PLD (14). Finally, studies in neutrophils (15) and monocytic U937 cells (16) have demonstrated that stimulation of Fc␥R is tightly linked to PLD activation.
In light of the growing body of evidence linking ARF activation to receptor stimulation of phagocytes, and findings linking PLD activation to ARF, as well as to receptor-mediated oxidative responses, we explored the possible involvement of ARF1 and ARF6 in NADPH oxidase activation and whether this involves participation of PLD. For this purpose, we used the PLB-985 cell line induced to differentiate into a neutrophil-like phenotype following treatment with dibutyryl cAMP (Bt 2 cAMP). In previous work (30) we demonstrated that these cells are readily transfected while exhibiting phenotypic traits of differentiated phagocytes. Using this model, we explored possible roles of ARF1 and ARF6 through transfection of mutated forms produced as fusions with enhanced green fluorescent protein (EGFP), and we demonstrated involvement of both ARF6 and PLD in formyl-methionine-leucine-phenylalanine (fMLP) receptor-mediated activation of the respiratory burst. In contrast, phorbol 12-myristate 13-acetate (PMA) activation of the oxidase appears to be PLD-dependent, but ARF independent, suggesting that ARF and protein kinase C act through different signaling pathways leading to oxidase activation, both apparently involving activation of PLD.

EXPERIMENTAL PROCEDURES
Cell Culture-PLB-985 cells were grown in stationary suspension cultures in RPMI 1640 medium containing 10% bovine serum (Hyclone Laboratories, Inc., Logan, UT), 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 12.5 units/ml nystatin at 37°C, in a humidified atmosphere of 5% CO 2 . Cell number and viability were determined by trypan blue exclusion.
Cell Labeling and NADPH Oxidase Activity Measurements by Flow Cytometry-Flow cytometric assays of NADPH oxidase activity involved the method from Model et al. (32) with slight modifications. After Ficoll gradient separation, the PLB-985 cells (10 7 cells/ml) were labeled in Ca 2ϩ -and Mg 2ϩ -free PBS containing 15 M 4-carboxydihydrotetramethylrosamine succinimidyl ester (Ros-SE; OxyBURST Orange-RE, Molecular probes, Eugene, OR), incubated on ice, and shaken in the dark for 30 min. After labeling, the cells were washed twice with cold Ca 2ϩ -and Mg 2ϩ -free PBS, resuspended at a concentration of 10 7 cells/ml in cold PBS, and kept on ice protected from light. The assay mixture for oxidase activity contained 100 l of ice-cold cell suspension (10 6 cells), 900 l of Hanks' balanced saline solution with Ca 2ϩ and Mg 2ϩ and 100 units/ml horseradish peroxidase (Sigma). The cells were prewarmed for 5 min in a 37°C shaking water bath followed by the addition of activators (1 M fMLP or 100 ng/ml PMA). Cells stimulated with fMLP were incubated for an additional 3 min; cells stimulated with PMA were incubated for an additional 15 min. After incubation the reactions were stopped by placing the tubes on ice and analyzed by flow cytometry within 1-2 h.
Flow Cytometry-Flow cytometric analysis was performed on 0.5 ϫ 10 6 cells using a Becton Dickinson FACStar Plus flow cytometer (San Jose, CA), equipped with an argon laser operating at 488 nm and spectra physics 2020 argon laser pumping a 355 dye laser head operating at 590 nm. Relative fluorescence signals were collected in fourdecade log 10 amplifiers and reported as linear geometric mean fluorescence value. To detect oxidized rosamine fluorescence the cells were excited at 488 nm, and the emission was detected with a 575/26 nm bandpass filter. EGFP fluorescence was detected with the same excitation wavelength using 512/20 nm bandpass filter. Because EGFP-transfected cells exhibited a broad range of EGFP expression levels, data collected from these cells were subdivided further into three regions based on fluorescence intensities: region 1 (R1, EGFP fluorescence ϭ 3 ϫ 10 1 -10 2 ); region 2 (R2, EGFP fluorescence ϭ 10 2 -10 3 ), and region 3 (R3, EGFP fluorescence ϭ 10 3 -10 4 ).
Detection of gp91phox by FACS Analysis-Transfected cells (10 6 ) were washed twice, resuspended in 100 l of Ca 2ϩ -and Mg 2ϩ -free PBS, and incubated for 30 min at room temperature with mouse anti-gp91phox antibody (7D5), which reacts with an extracellular epitope (33). The cells were then washed twice with PBS and incubated with goat anti-mouse IgG (HϩL) CY5-conjugated F(abЈ) 2 fragment (Jackson Immunoresearch Laboratories, Inc., West Grove, PA) for 30 min at room temperature in the dark. After washing, the cells were analyzed on the flow cytometer, and gp91phox was detected by excitation at 590 nm using a 668/14 bandpass filter. Transfected cells were detected independently by EGFP fluorescence, as described above. Histograms were constructed based on analysis of 0.5 ϫ 10 6 cells.
Chemiluminescence Assay of NADPH Oxidase Activity-Superoxide production by untransfected PLB-985 cells was assayed by chemiluminescence using a superoxide-specific, enhanced luminol-based substrate (DIOGENES, National Diagnostics), as described previously (34). The reactions were monitored for 15 min at 37°C following stimulation with 100 ng/ml PMA, using a Luminoskan plate-reading luminometer (Labsystem, Helsinki, Finland). Conditions for dose-and time-dependent inhibition of oxidase activity were explored with the PLD inhibitors 1-butanol and 2,3-diphosphogylcerate (2,3-DPG) (as well as 3-butanol, control) to determine the most effective treatment for inhibiting superoxide generation without reducing cell viability below 98%. Optimal concentrations were chosen at 0.5% 1-butanol, 0.5% 3-butanol, or 5 mM 2,3-DPG, which were added to the cells 10 min prior to activation.
Translocation and Immunoblot Analysis-Bt 2 cAMP-differentiated PLB-985 cells were stimulated for 3 min with 1 M fMLP or for 5 min with 100 ng/ml PMA at 37°C in Hanks' balanced saline solution with Ca 2ϩ and Mg 2ϩ . Cell membranes were prepared following sonication by methods described previously (30). 50 g of membrane proteins were separated by electrophoresis on 12% polyacrylamide SDS gels and blotted to nitrocellulose. ARF1 and ARF6 were detected by using rabbit anti-ARF1 (provided by Dr. Richard A. Kahn, Emory University, Atlanta, GA (35)) and rabbit anti-ARF6 (provided by Dr. Julie G. Donaldson, NIH, Bethesda, MD (26)) according to standard protocols (30).

RESULTS AND DISCUSSION
In previous studies we demonstrated the PLB-985 cell line to be a useful model system that is capable of developing a differentiated myeloid phenotype while being amenable to gene transfection protocols (30). Because of the relatively low transfection efficiency of these cells and the requirements for rapid transfection protocols, we developed a multicolor FACS analysis to identify and characterize small subpopulations of transiently transfected cells within mixed cell populations. Transfected cells were identified by detection of recombinant EGFP, expressed either alone or fused with various ARF proteins of interest. Following treatments with Bt 2 cAMP for 3 days to induce a neutrophil-like phenotype, the cells were analyzed for forward angle light scattering (FALS) and 90 0 side light scattering properties. Two distinct populations were observed based on distinct light scattering properties (Fig. 1A): a low FALS population, indicated as gate G1, and a high FALS population, indicated as gate G2 or G3 (corresponding to untransfected or transfected cell cultures, respectively). The high FALS population from transfected cell cultures (G3) exhibited higher 90 0 light scatter compared with control untransfected cells (G2), consistent with a greater granularity that was associated with either the electroporation or sedimentation proto-cols employed with these cultures. Fig. 1B (right panels) shows that the high FALS population from a transfected culture (G3) contained the majority of EGFP-positive transfected cells, whereas the low FALS cell population (G1) was not readily transfected, as indicted by the low EGFP fluorescence readings observed with this gated population. All subsequent fluorescence-based functional studies focused on the function of the G3 gated population exhibiting the highest level of EGFP fluorescence (emission wavelength ϭ 512 nm) as an indication of the highest levels of recombinant ARF expression (EGFP fluorescence greater than 10 3 ).
Flow cytometric methods were also used to monitor superoxide production by single cells by measuring changes in fluorescence (emitted at 575 nm) upon oxidation of rosamine conjugated to the cell surface. For this purpose, PLB-985 cells were prelabeled with Ros-SE just prior to stimulation. Surface-conjugated Ros-SE is oxidized directly by hydrogen peroxide in activated PLB-985 cells; this oxidation is apparently dependent on peroxidase activity because reactions that lacked exogenously added horseradish peroxidase showed no change in Ros-SE fluorescence (data not shown). As shown in Fig. 1C, activation of Bt 2 cAMP-differentiated PLB-985 cells with 1 M fMLP or with 100 ng/ml PMA caused significant increases in Ros-SE fluorescence compared with unstimulated cells. In further support of the notion that Ros-SE oxidation was an indirect assay of superoxide production, changes in Ros-SE fluorescence observed in cells stimulated by 1 M fMLP were completely inhibited by diphenyleneiodonium (8 M), a known flavoprotein inhibitor of NADPH oxidase activity (Fig. 1C) (36). Taken together, these results indicated that the oxidation of surface-bound Ros-SE reflected NADPH oxidase activation and that the accumulation of fluorescence under these conditions was not caused by auto-oxidation of the probe.
Having established compatible assays for both transfected EGFP expression and oxidase activation in single cells, we explored the possible involvement of transfected ARF proteins in NADPH oxidase activation. Bt 2 cAMP-differentiated PLB-985 cells were transiently transfected with WT forms of recombinant ARF1 or ARF6 produced as fusion proteins with EGFP (ARF1-WT, ARF6-WT) or with fusion proteins of two mutants of each ARF. ARF6(T27N) and ARF1(T31N) mutants represent putative dominant-negative mutants with reduced affinity for GTP, whereas the ARF6(Q67L) and ARF1Q71L) mutants represent putative active forms with reduced GTPase activity. Earlier work has shown that the fusion of these proteins at their C terminus with EGFP does not interfere with GTP-dependent cycling between Golgi membrane and cytoplasmic compartments (ARF1) or signaling through PLD in response to receptor stimulation (ARF1, ARF6) in whole transfected cells (37,38). We compared activation of NADPH oxidase in Bt 2 cAMP-differentiated PLB-985 cells, PLB-985 cells transfected with empty EGFP-N1 vector (control), and cells transfected with the various EGFP/ARF protein constructs in response to 1 M fMLP or 100 ng/ml PMA. Fig. 2A presents results from a representative double fluorescence FACS analysis of transfected, fMLP-activated PLB-985 cells. Cells expressing the highest levels of recombinant EGFP/ARF proteins in R3 exhibited the most dramatic results (Figs. 2B and 3). Superoxide production observed in differentiated PLB-985 cells transfected with ARF1-WT, the two ARF1 mutants, or ARF6-WT was similar to activity observed in the differentiated, untransfected parental line or differentiated cells transfected with empty EGFP-N1 vector. In contrast, cells transfected with ARF6 mutants showed dramatic alterations in oxidase activity. The PLB-985 cells transiently transfected with ARF6(Q67L) showed a significant elevation (p ϭ 0.03) in oxidase activity, whereas PLB-985 cells transfected with ARF6(T27N) did not generate any detectable superoxide in response to fMLP. These effects were only observed when using fMLP as an agonist; transfected cells stimulated with PMA, a nonphysiological activator presumed to bypass early receptoractivated signaling intermediates, showed no effects on oxidative output with the expression of the same recombinant ARF fusion proteins (data not shown). Superoxide production in response to fMLP was similar in all cell populations that ex- hibited low EGFP fluorescence (i.e. EGFP Fl Ͻ 30, R0) regardless of the ARF construct transfected (data not shown), indicating that all of the cultures analyzed had the same oxidative potential, and therefore the differences in fMLP-elicited oxidase activity observed were limited to cells expressing high levels of ARF6 mutants.
To confirm the relationship between ARF6 expression and oxidative output in fMLP-activated cells, we compared superoxide production in cells analyzed within several regions based on EGFP fluorescence intensities. Fig. 3 compares oxidase activity between PLB-985 cells transfected with ARF6 protein constructs (T27N, or Q67L, or WT) and control EGFP-N1 vector. The results extend those obtained in Fig. 2B by showing that transfection with ARF6-WT and EGFP-N1 had no effect on oxidase function in all regions analyzed, while confirming that the alterations in oxidase activity by transfected ARF6(T27N) or ARF6(Q67L) correlated closely with the amount of these proteins produced. A similar analysis of PLB-985 cells transfected with ARF1 protein constructs showed no effect on oxidase activity (data not shown).
To address concerns of whether differences in oxidase activity could be explained simply by differences in differentiation or expression of essential oxidase components, we examined the cell surface expression of gp91phox in these cultures by FACS analysis using a monoclonal antibody directed against an extracellular epitope of this protein (33). Fig. 4 shows that all transfected cell populations detected in R3 expressed comparable amounts of cell surface gp91phox, as indicated by peak levels of secondary anti-mouse antibody detected. These results showed that the alterations in oxidase activity observed with ARF6(T27N) or ARF6(Q67L) expression were not caused by differences in gp91phox expression and confirmed that all transfected cultures differentiated to a comparable extent, consistent with findings that demonstrated comparable oxidative output in all populations exhibiting low EGFP fluorescence. These findings provide further support to the conclusion that ARF6, but not ARF1, has a specific signaling role in fMLPmediated activation of the respiratory burst.
To explore further the notion that fMLP receptor-mediated activation of the respiratory burst involves ARF6 signaling through activation of PLD, we also examined mutant forms of ARF1 and ARF6 which are thought to be defective in their activation of PLD. A recent report (13) identified regions within the crystallographic structure of ARF1 important for hPLD1 activation. The substitution of asparagine to arginine at position 52 (N52R) completely abolished the ability of ARF1 to activate hPLD1 in vitro, while not affecting another ARF1associated response, the recruitment of coatamer to membranes. Based on this observation, we mutated asparagine 52 to arginine (N52R) in ARF1/EGFP, as well as the corresponding site within ARF6(N48R)/EGFP. As shown in Fig. 6, PLB-985 cells transfected with ARF1-WT or ARF1(N52R), as well as ARF6-WT, showed no effect on superoxide production, whereas cells transfected with ARF6(N48R) exhibited significant inhibition (p ϭ 0.01) of superoxide production. Although the effects of this mutation on PLD activation were not examined directly, these findings provide additional support to the notion that ARF6 participates in receptor-mediated oxidase activation and suggest that ARF6 acts through activation of PLD, consistent with results obtained with the pharmacological agents shown in Fig. 5.
Phorbol esters are also effective stimuli of PLD activity in a wide range of intact cells (41). In neutrophils and HL-60 cells, most agonists that activate the respiratory burst also activate protein kinase C and PLD (39). To clarify further whether protein kinase C-dependent stimulation of the oxidase involves PLD in Bt 2 cAMP-differentiated PLB-985 cells, we tested the effect of the same PLD inhibitors on PMA-elicited superoxide production in these cells (Fig. 7). PLD inhibitors, 0.5% 1-butanol or 5 mM 2,3-DPG, caused significant inhibition (p ϭ 0.005 and p ϭ 0.001, respectively) of superoxide production in response to PMA, whereas the control compound 3-butanol (0.5%) had no effect on superoxide. These observations, together with the absence of any demonstrable effects of dominant negative mutants of ARF1 or ARF6 on PMA-stimulated oxidase activity, suggest that protein kinase C activation of the oxidase is PLDdependent but ARF-independent.
As another correlate to ARF involvement in oxidase activation in differentiated PLB-985 cells, we examined membrane translocation of ARF1 and ARF6 following stimulation with either 1 M fMLP or 100 ng/ml PMA. Fig. 8 shows that stimulation by either of these agonists caused enhanced membrane binding of both the ARF1 and ARF6 isoforms. These observations are consistent with previous reports demonstrating that both fMLP and PMA stimulate translocation of ARF to the plasma membrane in HL-60 cells and neutrophils, although these studies did not distinguish between the two isoforms (18,19). Thus, PMA stimulates ARF1 and ARF6 activation and translocation to membranes in PLB-985 cells but has other direct or overriding effects on oxidase activation which appear to involve PLD but are insensitive to the effects of the dominant ARF1 or ARF6 isoforms.
The present study provides novel evidence for unique involvement of ARF6 in N-formyl peptide receptor-stimulated activation of NADPH oxidase because expression of the GTP-binding deficient mutant (ARF6(T27N)) inhibited superoxide production, whereas the GTPase-deficient mutant (ARF6(Q67L)) enhanced superoxide production. Furthermore, the effects of these ARF6 mutants on superoxide production were dose-dependent. In contrast, these dominant-negative and positive effects on superoxide production were not observed when the corresponding ARF1 proteins were expressed, indicating that this response is specific for ARF6. Furthermore, the inhibitory effects of the ARF6(N48R) mutant, as well as those of the PLD inhibitors 1-butanol and 2,3-DPG, provide support for a model in which ARF6 regulates the fMLP receptoractivated respiratory burst through a PLD-dependent mechanism. This model is consistent with longstanding observations correlating neutrophil receptor-mediated oxidative responses to elevations in phosphatidic acid levels (17,39). Recent reconstitution studies in HL60 cells showed that ARF1 activates hPLD1 and that both ARF1 and hPLD1 are involved in secretion of lysosomal granules (13). These observations, together with our findings, indicate that both ARF1 and ARF6 become activated in stimulated myeloid cells but that the two proteins are involved in different functions which may relate to their segregation into different cellular compartments (23).
In conclusion, we have developed a unique assay using molecular approaches to study the signaling cascade leading from N-formyl peptide receptor stimulation to NADPH oxidase activation in intact neutrophil-like cells. Using this assay we demonstrated that ARF6 has a direct role in NADPH oxidase regulation and suggest that this physiological function of ARF6 is mediated through PLD. ARF6 and protein kinase C appear to activate the oxidase in parallel pathways. PLD apparently participates in both pathways and is located downstream of either protein kinase C or ARF6. Future work should address the identity of the PLD isozyme that links ARF6 activation to the oxidase, as well as upstream signaling intermediates responsible for ARF6 activation.